rows_by_key works with pl.Array

[gab23r]

Description

I have this dataframe:

import polars as pl
df = (
    pl.DataFrame([
        pl.Series(name = "a", values = [[0,0], [0,1]], dtype=pl.Array(pl.Int64, 2)),
        pl.Series(name = "b", values = ["a", "b"])
    ])
)
# shape: (2, 2)
# ┌───────────────┬─────┐
# │ a             ┆ b   │
# │ ---           ┆ --- │
# │ array[i64, 2] ┆ str │
# ╞═══════════════╪═════╡
# │ [0, 0]        ┆ a   │
# │ [0, 1]        ┆ b   │
# └───────────────┴─────┘

And I want to transform this dataframe into this dictionnary:
{(0,0): "a", (0, 1): "b"}

The way to do it would be to used df.rows_by_key("a") but it fails with TypeError: unhashable type: 'list'
This is because pl.Array becomes list python object and so cannot be key of a dictionary. But I would be great is pl.Array get transform in tuple.

I think it make a lot of sense overall and this way we can translate any polars dataframe to any builtin python type:

• pl.List -> list
• pl.Struct -> dict
• pl.Array -> tuple

[gab23r]

I just realized that rows_by_key doesn't work with pl.Array but it takes a list of columns as a key so this workaround works fine and is even better:
python

df.select(c.a.arr.to_struct().struct.unnest(), "b").rows_by_key(("field_0", "field_1"), unique=True)
# {(0, 0): 'a', (0, 1): 'b'}